Author: Ada, Deep Tide TechFlow

After the market close on June 3rd Eastern Time, Broadcom announced its Q2 results for the fiscal year ending May 3, 2026. In absolute figures, this is a record quarterly report. Revenue was $22.19 billion, up 48% year-over-year, the highest quarterly growth rate for the company since January 2017; adjusted EPS was $2.44, exceeding analysts' consensus estimate of $2.40. However, the market was not focused on Q2 but on Broadcom's Q3 AI chip revenue guidance, which was $16 billion, up over 200% year-over-year, but nearly 7 percentage points lower than the consensus estimate of $17.2 billion from sell-side analysts. This gap, along with the company's slightly lower-than-expected software business, triggered a sharp response in the stock price.

Q2 Results Almost Perfect, AI Semiconductor Revenue Grows for 13 Consecutive Quarters

According to Broadcom's official disclosure, Q2 AI semiconductor revenue reached $10.8 billion, up 143% year-over-year, surpassing the company's previous guidance of $10.7 billion provided in March. CEO Chen Fuyang stated in the earnings report that this quarter's growth was driven by "the dual drivers of customized AI accelerators and AI network demand."

From a business perspective, semiconductor solutions revenue was $15.009 billion, up 79% year-over-year, accounting for 68% of total revenue; of this, AI semiconductor revenue accounted for 72%, while non-AI semiconductor revenue was $4.2 billion, up 6%, with a backlog of orders exceeding $6 billion, showing a cyclical recovery. Infrastructure software business (i.e. VMware) revenue was $7.178 billion, an increase of 9% year-over-year, consistent with the company's own guidance, but below the $7.32 billion analyst expectation from StreetAccount, with a gap of approximately $140 million.

Profitability also performed strongly. Adjusted EBITDA reached $15.2 billion, accounting for 69% of revenue, setting a new historical record; free cash flow was $10.26 billion, accounting for 46% of revenue; cash balance at the end of the quarter was $19.63 billion, up $5.4 billion quarter-over-quarter.

Q3 Guidance Revenue Exceeds Expectations, but AI Semiconductor Revenue "Is $1.2 Billion Short"

Broadcom's Q3 guidance for total revenue is $29.4 billion, up 84% year-over-year, higher than the consensus estimate of $28.54 billion from analysts; among which, the semiconductor revenue guidance is $20.5 billion, up 124% year-over-year. However, the AI semiconductor revenue guidance is $16 billion, which is 7% lower than the $17.2 billion consensus estimate collected by LSEG and other institutions; the gap is even larger compared to some more optimistic buy-side expectations.

More critically, Chen Fuyang did not raise the AI chip revenue guidance for the fiscal year 2026 during the conference call. According to CNBC, he reiterated during the call that "the company expects this momentum to continue into fiscal year 2027 and maintains the AI semiconductor revenue guidance of over $100 billion." Bernstein analyst Stacy Rasgon commented that it was the Q3 AI performance guidance that dragged down Broadcom's stock price.

Summing Q1 ($8.4 billion), Q2 ($10.8 billion) actual revenue and Q3, Q4 expectations, Broadcom's total AI chip sales for this fiscal year are expected to be around $56 billion, which is still about $1.6 billion short of analysts’ consensus expectation of $57.6 billion.

After-hours Drop of Over 13%, Options Market Had Already Priced in Volatility

Broadcom's stock price reacted sharply in after-hours trading. After the earnings report was released at 4 PM Eastern Time on June 3rd, AVGO initially fell about 5%; as the details of the guidance were disclosed during the conference call, the decline continued to widen, eventually falling over 15% at one point in after-hours trading, and finally closing down 13.78%. Calculating based on the closing price of about $479 before the earnings report, the single-day market value evaporated by over $270 billion.

It is noteworthy that capital markets had already been preparing for significant volatility following Broadcom's earnings report. According to several media reports, the options market had priced in Broadcom's single-day post-earnings volatility at around 7.8%, significantly higher than the historical average. This pricing reflects the dilemma faced by investors: Broadcom's stock price had rebounded over 60% from the March lows before the earnings season, and its increase this year has been nearly 40%, with a valuation (approximately 90 times earnings) far above the semiconductor industry average of around 69 times.

Due to this valuation concern, the market's implicit threshold for Broadcom's earnings report was set at "significantly exceeding expectations," and any guidance below a "blowout" performance could trigger profit-taking.

AI Network Revenue Share Expected to Fall from 40% to 30%

For the A-share optical module sector, Chen Fuyang's statement about the AI network business during the conference call may be more damaging than the overall AI guidance.

According to Yahoo Finance citing the conference call content, Chen Fuyang confirmed that this quarter, the AI network business accounted for "nearly 40%" of AI semiconductor revenue, but he also stated that this proportion is expected to "normalize over time to around 30%, rather than remaining near 40%."

This is the first time Broadcom's management has clearly provided a path for the decline in the proportion of AI network business. The AI network (including Ethernet switch chips, optical transceiver connection chips, and other links) is precisely the core revenue source related to Chinese A-share optical module leaders such as Innotron, Newisheng, and Tianfu Communication. All three companies have seen their stock prices rise significantly this year, with a combined market capitalization exceeding Moutai at one point, Innotron's dynamic P/E ratio is about 66 times, and Tianfu Communication reached 139 times, with the valuation assumptions built on expectations of continued high growth in AI networks.

Chen Fuyang's latest statement means that even if the demand for AI computing power remains high, the network segment's share may peak first. If this signal is accepted by buyers, the previous valuation premium of A-share optical module leaders will face direct testing.

Contagion Effect: Marvell Declines in After-hours Trading; Asian AI Supply Chain Under Pressure Today

The guidance effect from Broadcom has already begun to spill over. Marvell's stock price fell by about 9% in after-hours trading, narrowing to about 6% before this report was published; companies in the AI network/connection concept, such as Astera Labs and Credo Technology, have also been under pressure after-hours. It is worth mentioning that Marvell's stock soared 32% on June 2nd after Nvidia's CEO Jensen Huang called it "the next trillion-dollar company," and continued to rise 3.73% on June 3rd, but this later sell-off indicates that the previous day’s “Nvidia premium” is encountering concentrated profit-taking pressure.

For the Asian market, there are two core points of interest today. First, whether the A-share optical module leader "Yizhongtian" combination can digest Chen Fuyang's statement about the network revenue share decline; second, whether South Korea's SK Hynix, Samsung Electronics, and other HBM suppliers will be dragged down by the overall cooling of the AI narrative. Considering that on June 2nd, Innotron's trading volume exceeded half of the entire A-share optical module sector's daily trading volume, the sector's emotional reaction may be magnified.

However, the earnings report itself did not negate the long-term prosperity of AI computing power. Chen Fuyang again described the demand for AI chips as "difficult to meet" during the conference call and reiterated the goal of exceeding $100 billion in AI chip revenue for fiscal year 2027. UBS and other institutions had also entered a "buy the dip" logic after similar declines following Broadcom's earnings report last December. Whether this round of adjustment is a narrative turning point or just a routine profit-taking for overvalued stocks will need to be assessed based on subsequent conference calls from leading companies and the capital expenditure trends of large-scale cloud providers.